import java.io.File;
import java.io.FileInputStream;
-import java.io.FileOutputStream;
import java.io.InputStream;
-import java.io.PrintStream;
import org.apache.commons.httpclient.HttpClient;
+import org.apache.commons.httpclient.NameValuePair;
import org.dom4j.Element;
import org.wamblee.crawler.Action;
import org.wamblee.crawler.Configuration;
import org.wamblee.crawler.Crawler;
import org.wamblee.crawler.Page;
+import org.wamblee.crawler.PageException;
+import org.wamblee.xml.XslTransformer;
/*
* Copyright 2005 the original author or authors.
*/
/**
- * Entry point for the crawler.
+ * Test application which uses the crawler.
*/
-public class App {
+public final class App {
- private static final String LOG_FILE = "crawler.log";
+ /**
+ * Disabled constructor.
+ *
+ */
+ private App() {
+ // Empty
+ }
+
+ /**
+ * Runs a test program.
+ *
+ * @param aArgs
+ * Arguments. First argument is the crawler config file name and
+ * second argument is the start url.
+ * @throws Exception
+ * In case of problems.
+ */
+ public static void main(String[] aArgs) throws Exception {
+ String configFileName = aArgs[0];
+ String starturl = aArgs[1];
- public static void main(String[] args) throws Exception {
- String configFileName = args[0];
- String starturl = args[1];
-
- FileOutputStream fos = new FileOutputStream(new File(LOG_FILE));
- PrintStream os = new PrintStream(fos);
+ ConfigurationParser parser = new ConfigurationParser(new XslTransformer());
+ InputStream configFile = new FileInputStream(new File(configFileName));
+ Configuration config = parser.parse(configFile);
- try {
- ConfigurationParser parser = new ConfigurationParser(os);
- InputStream configFile = new FileInputStream(new File(
- configFileName));
- Configuration config = parser.parse(configFile);
+ HttpClient client = new HttpClient();
+ // client.getHostConfiguration().setProxy("localhost", 3128);
- HttpClient client = new HttpClient();
- // client.getHostConfiguration().setProxy("localhost", 3128);
-
- Crawler crawler = new CrawlerImpl(client, config);
+ Crawler crawler = new CrawlerImpl(client, config);
- System.out.println("Retrieving: " + starturl);
- Page page = crawler.getPage(starturl);
- showPage(page);
- page = page.getAction("channels-favorites").execute();
- recordInterestingShows(page);
- showPage(page);
- page = page.getAction("Nederland 1").execute();
- showPage(page);
- page = page.getAction("right-now").execute();
- showPage(page);
- page = page.getAction("Het elfde uur").execute();
- showPage(page);
- } finally {
- os.flush();
- os.close();
- System.out.println("Output written on '" + LOG_FILE + "'");
- }
+ System.out.println("Retrieving: " + starturl);
+ Page page = crawler.getPage(starturl, new NameValuePair[0]);
+ showPage(page);
+ page = page.getAction("channels-favorites").execute();
+ recordInterestingShows(page);
+ showPage(page);
+ page = page.getAction("Nederland 1").execute();
+ showPage(page);
+ page = page.getAction("right-now").execute();
+ showPage(page);
+ page = page.getAction("Het elfde uur").execute();
+ showPage(page);
}
/**
*/
private static void showPage(Page aPage) {
Action[] links = aPage.getActions();
- for (Action link: links) {
+ for (Action link : links) {
System.out.println("Link found '" + link.getName() + "'");
}
- Element element = aPage.getContent();
+ Element element = aPage.getContent();
System.out.println("Retrieved content: " + element.asXML());
}
-
- private static void recordInterestingShows(Page page) {
+
+ private static void recordInterestingShows(Page page) throws PageException {
Action[] channels = page.getActions();
- for (Action channel: channels) {
- examineChannel(channel.getName(), channel.execute().getAction("right-now").execute());
+ for (Action channel : channels) {
+ examineChannel(channel.getName(), channel.execute().getAction(
+ "right-now").execute());
}
}
-
- private static void examineChannel(String aChannel, Page aPage) {
- Action[] programs = aPage.getActions();
- for (Action program: programs) {
+
+ private static void examineChannel(String aChannel, Page aPage)
+ throws PageException {
+ Action[] programs = aPage.getActions();
+ for (Action program : programs) {
System.out.println(aChannel + " - " + program.getName());
- if ( program.getName().toLowerCase().matches(".*babe.*")) {
+ if (program.getName().toLowerCase().matches(".*babe.*")) {
Page programPage = program.execute();
- Action record = programPage.getAction("record");
+ Action record = programPage.getAction("record");
System.out.println("Recording possible: " + record != null);
}
}